From afff6c814b24911756f43a697c17b1d2146f292e Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Mon, 13 Feb 2012 10:15:30 +0530 Subject: [PATCH] Updated kurier --- recipes/kurier.recipe | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) diff --git a/recipes/kurier.recipe b/recipes/kurier.recipe index 21674f953f..8e300b68cd 100644 --- a/recipes/kurier.recipe +++ b/recipes/kurier.recipe @@ -13,9 +13,10 @@ class Kurier(BasicNewsRecipe): publisher = 'KURIER' category = 'news, politics, Austria' oldest_article = 2 - max_articles_per_feed = 200 + max_articles_per_feed = 100 + timeout = 30 + encoding = None no_stylesheets = True - encoding = 'cp1252' use_embedded_content = False language = 'de_AT' remove_empty_feeds = True @@ -29,9 +30,11 @@ class Kurier(BasicNewsRecipe): , 'language' : language } - remove_tags = [dict(attrs={'class':['functionsleiste','functions','social_positionierung','contenttabs','drucken','versenden','leserbrief','kommentieren','addthis_button']})] + remove_tags = [ dict(attrs={'id':['artikel_expand_symbol2','imgzoom_close2']}), + dict(attrs={'class':['linkextern','functionsleiste','functions','social_positionierung','contenttabs','drucken','versenden','leserbrief','kommentieren','addthis_button']}) + ] keep_only_tags = [dict(attrs={'id':'content'})] - remove_tags_after = dict(attrs={'id':'author'}) + remove_tags_after = [dict(attrs={'id':'author'})] remove_attributes = ['width','height'] feeds = [ @@ -41,7 +44,7 @@ class Kurier(BasicNewsRecipe): ,(u'Kultur' , u'http://kurier.at/rss/kultur_kultur_rss.xml' ) ,(u'Freizeit' , u'http://kurier.at/rss/freizeit_freizeit_rss.xml' ) ,(u'Wetter' , u'http://kurier.at/rss/oewetter_rss.xml' ) - ,(u'Verkehr' , u'http://kurier.at/rss/verkehr_rss.xml' ) + ,(u'Sport' , u'http://kurier.at/newsfeed/detail/sport_rss.xml' ) ] def preprocess_html(self, soup):